Part 4 of the TrustSense build arc. Start with part 1.
I’ve been building and breaking things online for a long time. Writing about it is how I process what I learn. This is that post for TrustSense v2.
The AI part isn’t the hard part anymore
In 2023, getting an AI model to analyze text for trustworthiness required a custom-built dataset, a trained model, and a team of people who knew what they were doing. I was the least technical person on that team and spent most of my time trying to understand what everyone else was building.
In 2026, I wrote a prompt on a Saturday afternoon and had something working by Sunday. The model — llama3.1:8b, free, runs locally — did in hours what took weeks before.
I want to be careful not to oversell this. The tool isn’t perfect. It gets confused by ambiguous cases. It sometimes flags things for the wrong reasons. But the baseline capability is genuinely there in a way it wasn’t three years ago, and that’s a meaningful shift.
The hard part is still communication
Getting the model to score a phishing email accurately: a few hours.
Getting the output to be useful to someone who doesn’t know what “source credibility” means: longer.
I rewrote the labels, the verdicts, the summary prompts, and the “what to do” section multiple times — not because the AI was getting things wrong, but because the way it was explaining things assumed too much. Every piece of text in that app went through the filter of: would the person who almost clicked that link understand this?
That’s not an AI problem. It’s a communication problem. It’s the same problem I’ve been working on in education for twenty years, just in a different context.
Local AI is real. Distribution is not solved.
This is the honest part.
TrustSense v2 runs entirely on your machine. Your text never goes anywhere. That privacy story is real and it matters — especially for a tool designed to handle sensitive messages.
But to run it right now, you need to:
- Have Python installed
- Create a virtual environment
- Install two libraries
- Download and install Ollama separately
- Pull a 5GB model
- Run a terminal command to start the app
That’s not a tool for most people. That’s a tool for people who are already comfortable with this stuff.
Local AI in 2026 is capable. It is not yet accessible. The gap between those two things is where the interesting work is happening, and it’s closing — but it’s not closed.
What I’m doing with this
This build is part of a practice I’m trying to establish: one working prototype per month, built in public, written about honestly. Not polished products. Working things. The prototype is not the point — the learning is.
TrustSense was the first one. Back at the CMU hackathon in 2023, I also worked on pitches for apps in education and medicine that didn’t advance. I haven’t dropped those ideas — they’re next in the queue. Building TrustSense v2 in a weekend, alone, with tools that didn’t exist three years ago, makes me more interested in going back to those problems, not less.
Each build gives me something concrete to share with the educators, families, and non-technical people I work with. It’s easy to talk about AI in the abstract. It’s more useful to say: here’s something I built, here’s how it works, here’s where it fell short.
If you’d want to use a version of TrustSense that runs in a browser — no installation, no terminal, just paste and check — reply and tell me. That’s what I’m thinking about for next month, if there’s enough interest. The privacy tradeoffs change, but the reach changes too.
The question I keep coming back to
The people most at risk from phishing, scams, and manipulative content are the least equipped to recognize it. The tools that could help require a level of technical comfort that most people don’t have.
That gap is real and I don’t have a clean answer for it. But I think building things — and writing honestly about what works and what doesn’t — is more useful than theorizing about it.
Source code is on GitHub: github.com/wiobyrne/trustsense-v2
Questions or corrections: hello@wiobyrne.com
TrustSense Build Arc: Part 1 · Part 2 · Part 3 · Part 4
More in the garden: Vibe Coding Monthly Prototypes · Trust and Sincerity Detection in AI · Digital Identity Trust and Security MOOC Framework